2013-05-25 I've started to read ``Text-oriented Software'' by Francesc Hervada-Sala. Francesc does not just tidy up surfaces but goes right to the core and tries to rethink the fundamental concepts of computing. There's high potential in his ideas. I'm much look- ing forward to his talk at ChaosSeminar in Ulm in October this year. Had a deeper look in to qpsmtpd. I like the software a lot. Sad- ly, the focus much to forward mail to qmail or postfix and treat general forwarding to an SMTP server only second class. This is bad, IMO. If you've read ESR's lessons of fetchmail, you should come to the conclusion, that qpsmtpd should *only* forward on way: to a generic SMTP server. This way, you'll cover all the MTAs in the same go. My investigations on spamassassin came to the conclusion that it's still black magic and I don't want to interfere with it much. I have it running, it does a fairly good job, and it needs more memory than the rest of the system. I could have a look at dspam, but is that's really better? Or I use bogofilter in the mail backend, too. But it would be nice to have more that just the bayes test and also to have a second spam checker in the backend that's different to the one I use on the frontend. Result: Situation not good but good enough and no time and few motivation to work on that stuff. Bad news: The server upgrade isn't fully done; one user told me he cannot send mails no more. I'm not sure what the problem real- ly is. The involved components should not have changed much. I'll try to solve it tomorrow. http://marmaro.de/lue/ markus schnalke